As South Africa counted the votes in its local government elections, a parallel battle was unfolding online—one that had little to do with the actual tally. Research revealed that the Electoral Commission of South Africa (IEC) had become a prime target for disinformation campaigns swirling across social media platforms. The narratives were familiar and dangerous: claims of vote-rigging, allegations that foreign nationals were being allowed to cast ballots, and videos presenting routine electoral procedures as evidence of ballot tampering. These stories spread rapidly during the election period, casting a shadow over the democratic process and raising concerns about the integrity of the institutions tasked with upholding it.
Kira Alberts, a research co-ordinator at Stellenbosch University’s Centre for Research on Democracy, offered a sobering assessment of the situation. Figures compiled by the IEC, Real411, and Moxii Africa pointed to approximately 700 reported stories or items of election-related content. But Alberts was quick to caution that this number was almost certainly just the tip of the iceberg. “The reality is this is incredibly difficult to determine,” she explained, adding that “700, I would guess, is quite an underreported number.” Researchers relied on members of the public reporting content and open-source tracking of platforms like X, TikTok, and Facebook. However, private messaging platforms—particularly WhatsApp—remained a significant blind spot. “WhatsApp is a closed platform. We can’t track that at all, although we know this is being heavily used,” Alberts said. Instead of trying to measure the exact reach of disinformation, researchers found it more useful to identify recurring narratives and understand the themes that were being pushed.
One of the most persistent targets was the IEC itself. Claims circulated that the commission was captured, corrupt, or working hand-in-glove with the African National Congress (ANC). Other narratives alleged that the elections were being rigged or that foreign nationals were being allowed to vote, particularly during the special voting period. Alberts was clear about the intent behind these messages: “We know this is to sow this distrust, to make sure that people don’t really believe in the elections and the institutions that are upholding our democracy.” Such narratives could discourage people from voting at all or influence how they voted by blurring the distinction between the IEC, the state, and the governing party. And as counting continued, these same claims could affect whether voters and political parties would accept the final results—a dangerous prospect for any democracy.
The situation was further complicated by the use of artificial intelligence to create or manipulate political material. Alberts cited an alleged AI-generated video depicting former president Nelson Mandela telling people to stop voting for the ANC, as well as an earlier video purporting to show Democratic Alliance (DA) figures discussing dealings with alleged criminal figure Vusimuzi ‘Cat’ Matlala. “The messages do pack a punch and get the message across whether or not the viewer is able to see if it is fake or not,” she said. The misleading content also included xenophobic and anti-immigration messaging circulated through networks of accounts. Tracing those responsible and establishing the content’s reach remained difficult, if not impossible. The figures supplied by Real411 for the period from October 4 to November 4 recorded 176 hate speech reports, 128 relating to incitement to violence, 126 classified as disinformation or misinformation, and 104 involving harassment. These figures add up to 534 reports across the four categories, though they shouldn’t be read as 534 unique cases, as the information supplied doesn’t establish whether complaints may have been recorded in more than one category.
Examples assessed by Real411 illustrated the problem vividly. One TikTok video showed IEC officials handling special-vote material, footage that was presented alongside claims of ballot-box stuffing and corruption. Real411 assessed the content as highly likely to constitute misinformation, noting that the footage showed routine electoral procedures and provided no evidence of ballot tampering or misconduct. The IEC, for its part, sought to strengthen its response through a partnership with Google, Meta, Microsoft, TikTok, and Moxii Africa. The agreement provided for direct communication with the platforms and allowed the commission to refer suspected disinformation for action under their own policies. The commission also gazetted a Disinformation Code on September 4, which forms part of the Electoral Code of Conduct and applies to political parties, candidates, and anyone acting on their behalf. The IEC designated Real411 to receive and assess public complaints in all 11 official languages, with the power to recommend that matters be referred to the commission or relevant platforms. The IEC could also refer complaints to the Electoral Court or issue public statements to correct misleading information.
Alberts described the collaboration with technology companies as “quite an encouraging move,” but she was under no illusions about the scale of the challenge. The task of monitoring content across an increasingly fragmented digital environment remained daunting. With platforms multiplying and messaging apps offering encrypted, private spaces where disinformation could spread unchecked, the battle against false narratives was far from won. The election may have been a snapshot in time, but the broader war over information—and trust in democratic institutions—was ongoing. For South Africa, as for many countries around the world, the fight against disinformation had become an essential part of protecting democracy itself.